CN113100781B - System and method for monitoring injury stimulus responsiveness in operation based on electroencephalogram coupling relation - Google Patents
System and method for monitoring injury stimulus responsiveness in operation based on electroencephalogram coupling relation Download PDFInfo
- Publication number
- CN113100781B CN113100781B CN202110381757.XA CN202110381757A CN113100781B CN 113100781 B CN113100781 B CN 113100781B CN 202110381757 A CN202110381757 A CN 202110381757A CN 113100781 B CN113100781 B CN 113100781B
- Authority
- CN
- China
- Prior art keywords
- electroencephalogram
- phase
- amplitude
- coupling
- data
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/48—Other medical applications
- A61B5/4821—Determining level or depth of anaesthesia
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7203—Signal processing specially adapted for physiological signals or for diagnostic purposes for noise prevention, reduction or removal
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7225—Details of analog processing, e.g. isolation amplifier, gain or sensitivity adjustment, filtering, baseline or drift compensation
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7235—Details of waveform analysis
- A61B5/725—Details of waveform analysis using specific filters therefor, e.g. Kalman or adaptive filters
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7235—Details of waveform analysis
- A61B5/7253—Details of waveform analysis characterised by using transforms
- A61B5/726—Details of waveform analysis characterised by using transforms using Wavelet transforms
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7235—Details of waveform analysis
- A61B5/7264—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems
Abstract
The invention discloses a system and a method for monitoring injury stimulus responsiveness in operation based on an electroencephalogram coupling relationship. Acquiring intraoperative multi-channel electroencephalogram data, and taking the reaction condition of a patient after injury stimulation as a label; extracting MI values in channels and among channels from the preprocessed multi-channel electroencephalogram data; constructing a data set by using the characteristics in the channel and among the channels with the labels, and dividing the data set into a training set and a testing set; and training the random forest classifier by using a training set, and verifying and optimizing the trained random forest classifier by using a testing set. The injury stimulation monitoring means based on the autonomic nerve is easily influenced by medicines and external factors, and the injury stimulation monitoring means based on the electroencephalogram nonlinear characteristics, such as entropy index, introduces the forehead muscle spectrum entropy, reduces the interpretability of indexes and is easily interfered by muscle relaxation medicines. The coupling phase amplitude mode adopted by the invention is an inherent state under anesthesia, the sensitivity of the index to the electromyographic signal is reduced, and the coupling phase amplitude mode has higher specificity.
Description
Technical Field
The invention relates to the field of sedation and analgesia monitoring in an operation, in particular to a system and a method for monitoring the injury stimulation responsiveness in the operation process by utilizing an electroencephalogram coupling relationship.
Background
The anesthesia monitoring in the operation mainly comprises two parts of sedation monitoring and analgesia monitoring, the sedation monitoring has more effective clinical means, and the analgesia monitoring means is far from insufficient. The key to analgesic monitoring is the monitoring of the responsiveness to the stimuli of the injury during surgery. There are two main types of methods for monitoring nociceptive responsiveness: one type monitors physiological characteristics affected by autonomic nerves based on autonomic nerve activity, and the method is easily interfered by factors such as drugs, diseases, environment and the like; the other type monitors the responsiveness of the noxious stimulus from the angles of frequency spectrum, nonlinearity and evoked brain electricity on the basis of brain electricity. However, the monitoring method based on electroencephalogram is limited by the defects that the relation between the frequency spectrum and the response of the noxious stimulus is not clear, the interpretability of nonlinear characteristics is not strong, the induced electroencephalogram needs to be superposed for many times, the anti-interference capability is weak and the like.
Anesthesia consciousness research has proved that the change of the brain electrical coupling mode is one of the main effects of anesthesia drugs on the brain consciousness state, but for the monitoring of the injury stimulation reactivity, a method for monitoring the injury stimulation reactivity based on the brain electrical coupling relationship is lacked at present. The method provided by the invention introduces the electroencephalogram coupling relationship into the monitoring of the injury stimulus in the operation, popularizes the mechanism research into the clinical monitoring, and can improve the specificity and detection precision of the monitoring of the injury stimulus in the operation.
Disclosure of Invention
The invention aims to overcome the defects of the prior art and provides an intra-operative injury stimulus responsiveness monitoring method based on an electroencephalogram coupling relationship.
An intra-operative injury stimulation responsiveness monitoring method based on electroencephalogram coupling relationship is characterized by comprising the following steps:
step (1), acquiring multichannel electroencephalogram data in an operation, and taking the reaction condition of a patient after injury stimulation (hereinafter referred to as injury stimulation reactivity) as a label;
the acquisition mode of the multi-channel electroencephalogram data is to select different channels of electroencephalogram data distributed in each area of the brain by adopting the international 10-20 system standard and referring to the double ears and mastoid bones.
The source of the noxious stimulation is laryngeal mask intubation or incisional stimulation;
the reaction of the patient after the injury stimulation refers to stress reaction caused by the injury stimulation, including the actions of the body or the head or the phenomena of swallowing, chewing, frown wrinkling and the like;
step (2) preprocessing the multichannel electroencephalogram data
Removing possible power frequency interference, electrocardio and electro-oculogram interference, myoelectricity artifact and baseline drift in the data by using FIR filter, independent component analysis, wavelet threshold denoising, quadratic variation difference reduction and other methods;
step (3), extracting MI values in channels and among channels from the preprocessed multi-channel electroencephalogram data; the method mainly comprises the steps of decomposing electroencephalogram signals, extracting electroencephalogram channel internal coupling characteristics and extracting electroencephalogram channel inter-coupling characteristics;
introducing a Modulation Index (MI) to depict a phase amplitude coupling mode in an electroencephalogram channel and between channels, and analyzing the relationship between the phase amplitude coupling mode in the electroencephalogram channel and between the channels and the responsiveness of noxious stimulation;
3-1 electroencephalogram signal decomposition
Based on the electroencephalogram data preprocessed in the step 2, decomposing the electroencephalogram data into K amplitude frequency segments and K phase frequency segments according to equally spaced center frequencies by adopting Morlet wavelet, wherein K is defined as 20 generally;
the Morlet wavelet equation is as follows:
wherein t and f are the current time point and the instantaneous frequency respectively; a ═ δtπ1/2)-1/2The energy is a normalization factor of the wavelet, so that the energy is not changed after the wavelet transformation; deltat＝1/2πδf，δtAnd deltafThe bandwidth adjustment factors of the time window and the frequency window which are wavelet functions respectively can pass through deltatAnd deltafThe time resolution and the frequency resolution of the wavelet are adjusted by parameters;
its center frequency f due to the characteristics of Morlet waveletOAnd deltafRatio f ofO/δfIs a constant value and has a bandwidth of 6 deltaf. The invention will fO/δfCenter frequency f set to 5, i.e. 1.2 times bandwidthO. The frequency signal decomposed by Morlet wavelet has the characteristics of small bandwidth of low frequency band and large bandwidth of high frequency band, and accords with the precondition of phase amplitude coupling detection.
3-2 extraction of features in electroencephalogram channels
3-2-1, combining the decomposed amplitude frequency section and the phase frequency section pairwise to calculate a phase amplitude modulation index MI, wherein the calculation method comprises the following steps:
where N is the equal length of the phase division of the phase frequency binsThe number of intervals (N is artificially defined and usually takes 20), P is the distribution function between the phase and amplitude of the signal, U is the distribution function assuming a uniform distribution of the signal, DkL(P, U) is a difference function for judging phase-amplitude distribution and uniform distribution by using Kullback-Leibler distance, DkLThe expression of (P, U) is:
wherein the distribution of U (j) is uniform, and U (j) is 1/N; p (j) is a histogram ratio occupied by the amplitude of the amplitude frequency section in the j-th interval after the phase of the phase frequency section is divided into N intervals;
3-2-2 removing false coupling of the phase amplitude modulation index MI obtained in the step 3-2-1 to obtain credible data; the method comprises the following specific steps:
randomly dividing the electroencephalogram data of different frequency bands, which are preprocessed in the step (2) and decomposed in the step (3-1), into two parts, exchanging the positions of the divided data, and repeating the operation for S times (S is artificially defined and is usually 200) to generate proxy data;
judging whether the MI value calculated according to the method in the step 3-2-1 for the electroencephalogram data of different frequency bands decomposed by the step 3-1 after the preprocessing in the step 2 is smaller than the phase amplitude modulation index MI of all proxy data of the corresponding decomposed frequency bandsurroThe average value +1.96 times of standard deviation, the preprocessed MI value is set to 0 as the MI value after removing the false couplingrmOtherwise, the MI value calculated by the preprocessed data is reserved; the process can be expressed as:
where Mean (MI)surro) Phase amplitude modulation index MI representing all proxy data of corresponding frequency bandsurroMean value of (1), std (MI)surro) Phase amplitude modulation index MI representing all proxy data of corresponding frequency bandsurroStandard deviation of (d);
MI for each electroencephalogram channelrmDrawing an MI common mode graph, and searching a coupling channel with the maximum difference before and after the noxious stimulation and a coupling frequency band corresponding to the coupling channel through pairing t test; obtaining MI under the channelrmThe values are used as features within the brain electrical channel.
Due to the local network characteristics of the brain, the channels screened by the method are generally concentrated in certain areas of the brain. The local network mode formed by the electroencephalogram channels can be preliminarily screened by the method. And the reduction of leads required by actual detection according to an analysis result can be achieved by matching with subsequent machine learning characteristic screening.
3-3 extraction of features between brain electrical channels
For different electroencephalogram channels, MI values of possible coupling modes among alpha frequency bands (8-13 HZ), theta frequency bands (4-8 HZ), delta frequency bands (1-4 HZ), beta frequency bands (14-26 HZ) and gamma frequency bands (26-44 HZ) are calculated respectively, and a formula (2) is adopted for calculation;
coupling modes which can exist among an alpha frequency band (8-13 HZ), a theta frequency band (4-8 HZ), a delta frequency band (1-4 HZ), a beta frequency band (14-26 HZ) and a gamma frequency band (26-44 HZ) comprise alpha phase-gamma amplitude, theta phase-gamma amplitude, delta phase-gamma amplitude, theta phase-beta amplitude and delta phase-beta amplitude coupling modes;
clustering all possible coupling mode MI values of all channels before occurrence of the noxious stimulation by adopting a k-means algorithm respectively; determining the category number k of the clusters according to an inflection point rule, selecting the category with the maximum MI value of the cluster center, and calculating the MI value of the basic phase amplitude coupling mode among the channels of the category;
carrying out pairing t test on MI values of each basic phase amplitude coupling mode in the calculated alpha phase-gamma amplitude, theta phase-gamma amplitude, delta phase-gamma amplitude, theta phase-beta amplitude and delta phase-beta amplitude, comparing the test result with a threshold value, if the test result is larger than the threshold value, carrying out pairing t test, and screening out a phase amplitude coupling mode with the maximum MI value difference, wherein the contrast label is the difference of the MI values between the channel pairs before and after the injury stimulation of the electroencephalogram data with the stress response caused by the injury stimulation; the MI value in the phase-amplitude coupling mode is acquired as the inter-channel characteristic.
Preferably, in the step 3-2 of extracting the features in the electroencephalogram channel, frontal lobe and forehead area channels are selected.
Preferably, in the step 3-3 of extracting the characteristics among the electroencephalogram channels, the coupling between the forehead channels and the frontal lobe channels and the coupling between the frontal lobe and the parietal lobe are selected.
Step (4), constructing a data set by using the characteristics in the channel and among the channels with the labels, and dividing the data set into a training set and a testing set;
and (5) training the random forest classifier by using the training set, and verifying and optimizing the trained random forest classifier by using the testing set.
The invention also aims to provide an intra-operative injury stimulus responsiveness monitoring system based on the electroencephalogram coupling relationship, which comprises a data acquisition module, a data preprocessing module, a feature extraction module and a model training module;
the data acquisition module is used for acquiring multichannel electroencephalogram data related to injury stimulation in an operation and then transmitting the electroencephalogram data to the data preprocessing module;
the data preprocessing module is used for preprocessing signals of the electroencephalogram data, and comprises but is not limited to a series of operations for improving signal-to-noise ratio of the signals, such as removal of electro-oculogram, myoelectricity artifact and baseline drift; then, the preprocessed electroencephalogram data are transmitted to a feature extraction module;
the characteristic extraction module is used for extracting features in an electroencephalogram signal channel and between channels and constructing a data set; then the data set is transmitted to a model training module;
the model training module trains a data set by adopting a machine learning method and constructs a random forest classifier model for judging and predicting the response of the noxious stimulus.
Furthermore, the method adopted by the data preprocessing module belongs to the mature technology mastered by the technical personnel in the field, and can effectively remove the noise possibly contained in the multichannel electroencephalogram signals.
Furthermore, the features extracted by the feature extraction module are phase amplitude coupling features of electroencephalogram signals in different frequency bands in a multi-channel electroencephalogram channel and among channels.
Furthermore, the model training module is used for further screening the features by utilizing a training set provided by the feature extraction module and constructing a model with the functions of judging and predicting the noxious stimulus responsiveness by combining with a machine learning algorithm. The model training module also comprises a method for verifying the effect of the model, and the effectiveness of the model is verified.
The invention has the beneficial effects that:
1. the method develops from frontal single-channel electroencephalogram injury stimulation evaluation to multi-channel electroencephalogram injury stimulation evaluation based on a local network coupling mode, improves interpretability of the evaluation method and discrimination and prediction capabilities of injury stimulation reaction, and can obtain higher discrimination and prediction accuracy.
2. Modulation Index (MI) was introduced as an indicator for monitoring noxious stimulus responsiveness and further confirmed its effectiveness. The MI-characterized phase-amplitude coupling mode is mainly focused on certain areas of the brain, and there will be guidance to further reduce leads, enabling a combination of mechanistic and clinical applications.
3. The injury stimulation monitoring means based on the autonomic nerve is easily influenced by medicines and external factors, and the injury stimulation monitoring means based on the electroencephalogram nonlinear characteristics, such as entropy index, introduces the forehead muscle spectrum entropy, reduces the interpretability of indexes and is easily interfered by muscle relaxation medicines. The coupling phase amplitude mode adopted by the invention is an inherent state under anesthesia, the sensitivity of the index to the electromyographic signal is reduced, and the coupling phase amplitude mode has higher specificity.
Drawings
FIG. 1 is an overall flow chart of the present invention;
FIG. 2 is a schematic illustration of MI calculation and spurious coupling removal;
FIG. 3 is an implementation specific data channel profile;
FIG. 4 is a flow chart of an experiment in practice;
FIG. 5 is a phase-amplitude coupled common mode plot within a channel;
FIG. 6 is ANOVA analysis statistics before and after nociceptive stimulation;
FIG. 7 is a channel distribution of K-means versus pre-noxious-stimulation MI clustering;
FIG. 8 is an inter-channel phase-amplitude coupling connection pattern;
FIG. 9 is a noxious stimulus responsiveness discriminating ROC curve;
FIG. 10 is a ROC curve for the prediction of nociceptive responsiveness.
Detailed Description
The above-mentioned aspects of the present invention are further illustrated by the following specific examples. It should not be understood that the scope of the above-described subject matter of the present invention is limited to the following examples. Various substitutions and alterations according to the general knowledge and conventional practice in the art are intended to be included within the scope of the present invention without departing from the technical spirit of the present invention as described above.
The method for monitoring the responsiveness of injury stimuli in operation based on electroencephalogram coupling relationship comprises the following steps as shown in figure 1:
(1) experimental design and data acquisition
The source of noxious stimulation in this example utilizes strong direct current simulated incisional stimulation.
After approval by the ethical committee of the first subsidiary hospital of Zhejiang university and written informed consent of all the participating patients, 26 phase selective lower limb surgery patients without psychiatric history and chronic pain (grade ASA 1 or grade 2, 13 of men; age 18-65 years) were recruited. The patient wears the multichannel electroencephalogram cap and acquires electroencephalogram data of 20 channels of the patient, and the acquired channels are shown in figure 3. When the patient is in a resting state, the patient is subjected to combined anesthesia of lumbar anesthesia and epidural anesthesia. Anesthesia induction and maintenance are carried out by using sevoflurane, and the concentration of the sevoflurane is increased by 0.2% every 1-2min until eyelash reflex is lost. The concentration of sevoflurane is increased to 5% for at least 5min, then a laryngeal mask is intubated, and maintenance is performed by reducing the concentration of sevoflurane to 2% after the patient stabilizes. In the latter half of the operation, the concentration of sevoflurane is kept at 2.5% for 10min, the median nerve of the hand of the patient is stimulated by nociceptive current (10mA,10s,50HZ or 40mA,10s,50HZ), the reaction within 1-2min after the electrical stimulation is observed, and the body movement condition of the patient is recorded. After electrically stimulating for 2-5min, regulating concentration of sevoflurane to 2%, maintaining for 10min, performing strong direct electrical stimulation on median nerve of hand of patient, and observing reaction condition of patient within 1-2min after stimulation. The same procedure was carried out at a concentration of 1.5% sevoflurane, except that the noxious stimulation was carried out at 2% and 2.5% sevoflurane concentrations. After the operation is finished, the concentration of the sevoflurane is gradually reduced, and the patient is waited for revival. Figure 4 shows the experimental procedure when embodied.
(2) Data pre-processing
The invention adopts an FIR notch filter to filter 50HZ power frequency interference, and simultaneously adopts Independent Component Analysis (ICA) to remove electrocardio and electro-oculogram interference possibly existing in electroencephalogram signals. And the noise with wide frequency band distribution, such as myoelectric interference, is removed by adopting a method of combining an FIR low-pass filter with wavelet threshold denoising. In addition, for baseline drift, the invention adopts a quadratic variation difference reduction method to estimate the baseline and remove the baseline.
(3) Analysis of coupling relationship and noxious stimulus responsiveness
Phase amplitude coupling in channels:
in order to accurately lock the frequency pair of phase-amplitude coupling change under the action of the noxious stimulation, all channel electroencephalogram signals before and after the noxious stimulation within 2min are analyzed. Selecting a 30s time window, and dividing the time sequence of each channel into two groups of frequency groups by using the Morlet wavelet in the step 3 in the time window: a set of phase frequencies and a set of amplitude frequencies. Wherein the phase frequency group consists of 20 frequency segments, the frequency range is 0-15HZ, and the frequency center is divided at equal intervals of the frequency range; the amplitude frequency group consists of 20 frequency segments, the frequency range is 0-50HZ, and the frequency center is divided at equal intervals of the frequency range. After calculating the MI index pairwise between the phase frequency segment group and the amplitude frequency group in the step 3-2-1, the false coupling is removed by the proxy data method in the step 3-2-2 (fig. 2).
The phase-amplitude coupling of each channel was plotted and the significance of coupling changes before and after stimulation was examined by a method of paired ANOVA analysis. The three channels with the strongest phase-amplitude coupling were examined to be Fpz, Fz, T8, and the phase-amplitude common mode diagram is shown in FIG. 5.
Patients were divided into physical (i.e., stressed) and anemic (i.e., unstressed) groups according to whether the patients were stressed after the noxious stimulation. ANOVA analysis of variance is carried out on MI values before and after the injury stimulation, and the MI values of theta phase-gamma amplitude of frontal lobe and frontal lobe of the patient in the body movement group are found to have significant difference. As shown in fig. 6, MI values for the frontal and frontal channels θ - γ decreased significantly when the nociceptive stimulus caused the patient to move.
② phase amplitude coupling between channels:
according to the step 3-3, the method of combining kmeans and quartile is adopted to determine the basic phase amplitude coupling mode before the occurrence of the noxious stimulus, and the coupling frequency pair most relevant to the responsiveness of the noxious stimulus, namely the phase amplitude coupling mode with the maximum MI value difference, is determined by adopting the paired t test, and finally 68 channel pairs of theta-gamma coupling are reserved as the research objects. Fig. 7 shows the results of using kmeans to classify the coupling between channels into two classes, while fig. 8 shows the phase-amplitude coupling pattern between channel pairs before and after the noxious stimulation. The results show that before the onset of the nociceptive stimulation, whether in the body movement group or the anemic group, the phase-amplitude coupling of θ - γ is mainly concentrated between the forehead and the frontal lobe channels, and the phase-amplitude modulation mode is mostly the amplitude of the posterior channel to modulate the phase of the anterior channel. Phase-amplitude coupling between frontal and parietal channels in the somatotrophic group is more pronounced before nociceptive stimulation is initiated. After the nociceptive stimulation, the phase-amplitude coupling connection between the forehead and frontal lobe channels of the physical exercise group is weakened, while the phase-amplitude coupling between the forehead and frontal lobe channels of the physical exercise group is substantially preserved. The coupling between the forehead, the frontal lobe passageways, and the frontal and parietal lobes may reflect the ability of the patient to respond to noxious stimuli.
(4) Constructing model datasets
The collected clinical data of 26 subjects were divided into 2 sub-data sets, i.e., an immotile group and a body movement group, according to whether the subjects were physically moved or not. The problem of monitoring the responsiveness of the noxious stimulus can be converted into 2 binary problems, one is to judge whether the current state is the response to the noxious stimulus; another is to predict whether a response will occur if a noxious stimulus is given in the current state. For the problem of judging the responsiveness of the noxious stimulus, the data labels of all the data sections before and after the noxious stimulus of the body motion free group are marked as 0, the data before the noxious stimulus of the body motion group occurs is marked as 0, and the data after the noxious stimulus occurs is marked as 1; for the problem of the response prediction of the noxious stimulation, data of a patient using 40mA tetanic electrical stimulation before the occurrence of the noxious stimulation is labeled, data of an anemic group before the occurrence of the noxious stimulation is labeled as 0, and data of an athletic group before the occurrence of the noxious stimulation is labeled as 1.
After the data is labeled, calculating the phase amplitude coupling characteristics MI in the channels and among the channels of the multi-channel electroencephalogram. And calculating the characteristic value by adopting a mode of sliding a 30s time window, wherein the sliding step length is 1 s.
And intercepting the electroencephalogram data within 1min before and after the injury stimulation to calculate the characteristics, stopping sliding when the sliding window exceeds the intercepted time window, and inheriting the label of the data section adopted by calculation by the characteristic value.
(5) Noxious stimulus responsiveness distinguishing and predicting model
Aiming at the problem of monitoring the responsiveness of the injury stimulus in the operation, the invention combines the phase amplitude coupling characteristics by using a random forest algorithm to construct a model for evaluating the responsiveness of the injury stimulus. The main tasks of the model are two: 1. judging whether stress reaction occurs after the noxious stimulation; 2. and classifying data before the nociceptive stimulation, and estimating whether body motion occurs after 40mA strong direct electrical stimulation.
For the evaluation of the model, the accuracy, the F1 measurement, the area under the ROC curve (AUC) and the PK prediction probability are adopted, and the final evaluation results are the average of the evaluation results after 10-fold cross validation. The PK prediction probability is not commonly used in a machine learning algorithm and is a method for evaluating indexes commonly used in the field of anesthesia, and under a general condition, a monotonous non-decreasing mathematical relationship is presented between the anesthesia depth and the anesthesia depth indexes, so that the PK prediction probability is more suitable for monitoring the anesthesia depth. PK 1 indicates that the change of the label can be completely reflected, PK 0.5 indicates that the result is almost randomly guessed for the label. PK<At 0.5, the results are shown to be in contrast to the label change, when 1-PK was used as the estimated PK prediction probability. PK value measurementCalculation statistical software SPSS calculates d proposed by Kim and the likeyxReuse of (1+ d)yx) The PK prediction probability is estimated.
Fig. 9 and 10 show ROC curves in 10-fold cross validation for the nociceptive responsiveness discrimination task and the nociceptive responsiveness prediction task, respectively. Table 1 and table 2 show the model results of each evaluation index evaluation.
TABLE 1 evaluation of discriminant models of nociceptive response
TABLE 2 evaluation of predictive models of nociceptive response
The above embodiments are not intended to limit the present invention, and the present invention is not limited to the above embodiments, and all embodiments are within the scope of the present invention as long as the requirements of the present invention are met.
Claims (9)
1. An intra-operative injury stimulation responsiveness monitoring method based on electroencephalogram coupling relationship is characterized by comprising the following steps:
step (1), acquiring multichannel electroencephalogram data in an operation, and taking the reaction condition of a patient after injury stimulation as a label;
step (2), preprocessing the multichannel electroencephalogram data;
step (3), extracting MI values in channels and among channels from the preprocessed multi-channel electroencephalogram data; the method mainly comprises the steps of decomposing electroencephalogram signals, extracting electroencephalogram channel internal coupling characteristics and extracting electroencephalogram channel inter-coupling characteristics;
3-1, carrying out electroencephalogram signal decomposition on the electroencephalogram data preprocessed in the step (2)
3-2 extraction of features in electroencephalogram channels
3-2-1, combining the decomposed amplitude frequency section and the phase frequency section pairwise to calculate a phase amplitude modulation index MI;
3-2-2 decouples the phase amplitude modulation index MI obtained in step 3-2-1 to obtain reliable data
；
For each channel of the brain electricity
Drawing an MI common mode graph, and searching a coupling channel with the maximum difference before and after the noxious stimulation and a coupling frequency band corresponding to the coupling channel through pairing t test; get under the channel
The value is used as the characteristic in the brain electricity channel;
3-3 extraction of features between brain electrical channels
Respectively calculating MI values of possible coupling modes among the alpha frequency band 8-13 HZ, the theta frequency band 4-8 HZ, the delta frequency band 1-4 HZ, the beta frequency band 14-26 HZ and the gamma frequency band 26-44 HZ among different electroencephalogram channels; wherein there may be coupling modes including alpha phase-gamma amplitude, theta phase-gamma amplitude, delta phase-gamma amplitude, theta phase-beta amplitude, and delta phase-beta amplitude coupling modes;
clustering all possible coupling mode MI values of all channels before occurrence of the noxious stimulation; selecting the category with the maximum MI value of the clustering center, and calculating the MI value of the basic phase amplitude coupling mode among the channels of the category;
carrying out pairing t test on MI values of each basic phase amplitude coupling mode in the calculated alpha phase-gamma amplitude, theta phase-gamma amplitude, delta phase-gamma amplitude, theta phase-beta amplitude and delta phase-beta amplitude, comparing the test result with a threshold value, if the test result is larger than the threshold value, carrying out pairing t test, and screening out a phase amplitude coupling mode with the maximum MI value difference, wherein the contrast label is the difference of the MI values between the channel pairs before and after the injury stimulation of the electroencephalogram data with the stress response caused by the injury stimulation; obtaining an MI value in the phase amplitude coupling mode as an inter-channel characteristic;
step (4), constructing a data set by using the characteristics in the channel and among the channels with the labels, and dividing the data set into a training set and a testing set;
and (5) training the random forest classifier by using the training set, and verifying and optimizing the trained random forest classifier by using the testing set.
2. The method for monitoring the responsiveness of the injury stimulus during the operation based on the electroencephalogram coupling relationship as claimed in claim 1, wherein the source of the injury stimulus is laryngeal mask intubation or incisional skin stimulus.
3. The method for monitoring the responsiveness of the injury stimulus in the operation based on the electroencephalogram coupling relationship as claimed in claim 1, wherein the step (2) is specifically to remove possible power frequency interference, electrocardio-electro-oculogram interference, electromyogram artifact and baseline drift in the data by using an FIR filter, independent component analysis, wavelet threshold denoising and secondary variation reduction method.
4. The method for monitoring the responsiveness of the injury stimulus in the operation based on the electroencephalogram coupling relationship as claimed in claim 1, wherein Morlet wavelet is adopted in step 3-1 to decompose the electroencephalogram data preprocessed in step (2) into K amplitude frequency segments and K phase frequency segments according to the center frequency at equal intervals, wherein K is a defined parameter; the Morlet wavelet equation is as follows:
wherein t and f are the current time point and the instantaneous frequency respectively;
normalized factor of wavelet;
，
and
respectively, the time window and frequency window bandwidth adjustment factors of the wavelet function.
5. The method for monitoring the responsiveness of the injury stimulus in the operation based on the electroencephalogram coupling relationship as claimed in claim 1, wherein the step 3-2-2 is as follows:
randomly dividing the electroencephalogram data preprocessed in the step (2) into two parts, exchanging the positions of the divided data, and repeating the operation S times, wherein S is defined parameters and generates proxy data;
judging whether the MI value calculated by the electroencephalogram data of different frequency bands which are preprocessed in the step (2) and decomposed in the step 3-1 according to the method in the step 3-2-1 is smaller than the phase amplitude modulation index of all proxy data of the corresponding decomposition frequency band
The average value +1.96 times the standard deviation, the preprocessed MI value is set to 0 as the value after removing the false coupling
Value of
Otherwise, the MI value calculated by the preprocessed data is reserved; the process can be expressed as:
wherein
Phase amplitude modulation index representing all proxy data
The average value of (a) of (b),
phase amplitude modulation index representing all proxy data
Standard deviation of (2).
6. The method for monitoring the responsiveness of injury stimuli in operation based on electroencephalogram coupling relationship according to claim 1, characterized in that the clustering in the step 3-3 adopts a k-means algorithm; the number k of the clusters is determined according to the inflection point rule.
7. The method for monitoring the responsiveness of the injury stimuli in the operation based on the electroencephalogram coupling relationship as claimed in claim 1, wherein the frontal lobe and frontal area channels are selected in the process of extracting the characteristics in the electroencephalogram channels in the step 3-2.
8. The method for monitoring the responsiveness of the injury stimuli in the operation based on the electroencephalogram coupling relationship as claimed in claim 1, wherein the coupling between the channels of the forehead and the frontal lobe and the coupling between the frontal lobe and the parietal lobe are selected in the extraction process of the characteristics among the electroencephalogram channels in the step 3-3.
9. The method for monitoring the responsiveness of the injury stimulus in operation based on the electroencephalogram coupling relationship as claimed in claim 1, wherein the phase amplitude modulation index MI in the step 3-2-1 is calculated by the following steps:
where N is the number of equal-length intervals of the phase division of the phase frequency bin, P is the distribution function between the phase and amplitude of the signal, and U is the assumed signal uniform divisionThe distribution function of the cloth time is that,
in order to judge the difference function between the phase-amplitude distribution and the uniform distribution by using the Kullback-Leibler distance,
the expression of (a) is:
wherein
Is uniformly distributed, and is taken
；
The phase of the phase frequency segment is divided into N intervals, and the amplitude of the amplitude frequency segment in the j-th interval occupies histogram ratio.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110381757.XA CN113100781B (en) | 2021-04-09 | 2021-04-09 | System and method for monitoring injury stimulus responsiveness in operation based on electroencephalogram coupling relation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110381757.XA CN113100781B (en) | 2021-04-09 | 2021-04-09 | System and method for monitoring injury stimulus responsiveness in operation based on electroencephalogram coupling relation |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113100781A CN113100781A (en) | 2021-07-13 |
| CN113100781B true CN113100781B (en) | 2022-02-18 |
Family
ID=76715228
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110381757.XA Active CN113100781B (en) | 2021-04-09 | 2021-04-09 | System and method for monitoring injury stimulus responsiveness in operation based on electroencephalogram coupling relation |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113100781B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114343635A (en) * | 2021-12-06 | 2022-04-15 | 北京理工大学 | Variable phase-splitting amplitude coupling-based emotion recognition method and device |
| CN114081494B (en) * | 2022-01-21 | 2022-05-06 | 浙江大学 | Depression state detecting system based on brain lateral reins signal |
| CN116011506B (en) * | 2023-03-28 | 2023-06-09 | 同心智医科技(北京)有限公司 | Method for constructing stereo electroencephalogram electrode signal decoding model and application thereof |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109875507A (en) * | 2019-01-28 | 2019-06-14 | 北京航空航天大学 | A kind of positioning of the area Zhi Xian and the potential area Zhi Xian prediction technique based on phase width coupling network |
| CN109893097A (en) * | 2019-03-28 | 2019-06-18 | 西安交通大学 | A kind of anesthesia depth monitoring system and method based on the coupling of near-infrared phase-amplitude |
| CN110495880A (en) * | 2019-08-16 | 2019-11-26 | 杭州电子科技大学 | Dyskinesia cortex plasticity management method based on the coupling of electrical transcranial stimulation brain flesh |
| CN111110231A (en) * | 2020-01-15 | 2020-05-08 | 燕山大学 | Consciousness assessment method based on multichannel neural signal coupling analysis |
| CN111513704A (en) * | 2020-06-09 | 2020-08-11 | 首都医科大学附属北京天坛医院 | Index and system for predicting prognosis of acute ischemic stroke patient receiving revascularization treatment |
| CN111712193A (en) * | 2018-02-14 | 2020-09-25 | 塞勒尼恩有限公司 | Apparatus and method for electroencephalography measurements |
| CN113274033A (en) * | 2021-05-10 | 2021-08-20 | 燕山大学 | Movement function monitoring and management method based on cross frequency coupling of brain and muscle electricity |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2016074103A1 (en) * | 2014-11-14 | 2016-05-19 | Neurochip Corporation | Method and apparatus for processing electroencephalogram (eeg) signals |
-
2021
- 2021-04-09 CN CN202110381757.XA patent/CN113100781B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111712193A (en) * | 2018-02-14 | 2020-09-25 | 塞勒尼恩有限公司 | Apparatus and method for electroencephalography measurements |
| CN109875507A (en) * | 2019-01-28 | 2019-06-14 | 北京航空航天大学 | A kind of positioning of the area Zhi Xian and the potential area Zhi Xian prediction technique based on phase width coupling network |
| CN109893097A (en) * | 2019-03-28 | 2019-06-18 | 西安交通大学 | A kind of anesthesia depth monitoring system and method based on the coupling of near-infrared phase-amplitude |
| CN110495880A (en) * | 2019-08-16 | 2019-11-26 | 杭州电子科技大学 | Dyskinesia cortex plasticity management method based on the coupling of electrical transcranial stimulation brain flesh |
| CN111110231A (en) * | 2020-01-15 | 2020-05-08 | 燕山大学 | Consciousness assessment method based on multichannel neural signal coupling analysis |
| CN111513704A (en) * | 2020-06-09 | 2020-08-11 | 首都医科大学附属北京天坛医院 | Index and system for predicting prognosis of acute ischemic stroke patient receiving revascularization treatment |
| CN113274033A (en) * | 2021-05-10 | 2021-08-20 | 燕山大学 | Movement function monitoring and management method based on cross frequency coupling of brain and muscle electricity |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113100781A (en) | 2021-07-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113100781B (en) | System and method for monitoring injury stimulus responsiveness in operation based on electroencephalogram coupling relation | |
| CN109770924B (en) | Fatigue classification method for building brain function network and related vector machine based on generalized consistency | |
| Wilson et al. | Spike detection: a review and comparison of algorithms | |
| EP2498676B1 (en) | Brain activity as a marker of disease | |
| Chen et al. | Automatic sleep stage classification based on subthalamic local field potentials | |
| Niederhauser et al. | Detection of seizure precursors from depth-EEG using a sign periodogram transform | |
| US20150245800A1 (en) | Method for Detection Of An Abnormal Sleep Pattern In A Person | |
| KR20110094064A (en) | Methods of identifying sleep &waking patterns and uses | |
| US20180279938A1 (en) | Method of diagnosing dementia and apparatus for performing the same | |
| CN112057087B (en) | Autonomic nerve function data processing method and device for high-risk schizophrenic people | |
| CN112426162A (en) | Fatigue detection method based on electroencephalogram signal rhythm entropy | |
| Fürbass et al. | Combining time series and frequency domain analysis for a automatic seizure detection | |
| Höller et al. | Do eeg-biometric templates threaten user privacy? | |
| Qi et al. | Driver fatigue assessment based on the feature fusion and transfer learning of EEG and EMG | |
| Wang et al. | Application of fast perturbational complexity index to the diagnosis and prognosis for disorders of consciousness | |
| Liang et al. | Propofol anesthesia decreased the efficiency of long-range cortical interaction in humans | |
| Wei et al. | Development of a rule-based automatic five-sleep-stage scoring method for rats | |
| Fathima et al. | Wavelet based features for classification of normal, ictal and interictal EEG signals | |
| CN115429292A (en) | Electroencephalogram signal quality detection device and system based on spectrum analysis | |
| CN113040788A (en) | Electroencephalogram signal quality detection method based on spectrum analysis | |
| Demirel et al. | Single-channel EEG based arousal level estimation using multitaper spectrum estimation at low-power wearable devices | |
| Chibirova et al. | Unsupervised Spike Sorting of extracellular electrophysiological recording in subthalamic nucleus of Parkinsonian patients | |
| Zhang et al. | EEG signal classification method based on improved empirical mode decomposition and SVM | |
| Chen et al. | Automatic multichannel intramuscular electromyogram decomposition: progressive fastICA peel-off and performance validation | |
| Cai et al. | Analysis of spontaneous EEG activity in Alzheimer’s disease using weighted visibility graph |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |